What History Tells Us About the AI Revolution | Tyler Cowen & Alex Imas

发布时间    来源
Episode 设置


登录已过期或未登录,无法修改。请先登录后再试。

以下是该播客内容摘要的中文翻译: 以下是该播客的摘要,包含了文稿中提出的所有新闻、事实和观点: ### **关于人工智能的总体情绪与经济前景** * **人工智能的格局与进展:** 这场对话发生于2026年11月11日。亚历克斯(Alex)和泰勒(Tyler)都认为,人工智能的进展令人难以置信且正在加速,其标志是模型发布频繁且功能更加强大。 * **聊天机器人时代与智能体时代:** 从聊天机器人时代(被用作高级维基百科)向智能体时代(系统执行实际工作)的转变,导致了能力和Token使用量的巨大加速。然而,智能体系统消耗的Token数量要多得多,前沿实验室(OpenAI、Anthropic、Muse、DeepMind)之间的价格竞争推低了Token价格,尽管Token使用量呈指数级增长,但营收数字却持平。 * **经济影响与增长率:** * 泰勒和亚历克斯反驳了关于失控的双位数GDP增长(奇点式增长)的“旧金山共识”。根据数据以及本·琼斯(Ben Jones)、查德·琼斯(Chad Jones)和阿吉翁(Aghion)关于理念生产函数的论文,收益递增的参数绝对小于1,这指向收益递减和更正常的个位数增长路径。 * 市场数据(实际利率、股票价格)、Anthropic首次公开募股(IPO)的预测市场(没有飞涨超过2万亿美元)以及正常的机构运营,表明经济正处于更正常的路径上。 * 真正的经济转型不仅需要能力,还需要智能体相互协调的能力。目前,智能体集群不擅长协调,主要进行并行处理。 * 美国经济包含许多巨大的瓶颈、惯性和非初创行业(医疗保健、高等教育、政府、非营利组织),这些都减缓了技术的普及。 * 泰勒估计,智能体达到占GDP的10%这一目标不会很快发生(尽管在五年内达到1.5%是可能的)。 * **宏观与政治事件:** * 中期选举使泰勒更新了对得克萨斯州的看法,该州出现反转,这让他感到惊讶。 * 近期的选举以及围绕数据中心的政治摩擦可能会由于监管而略微增加成本并放慢进度,但不会永远停滞该行业。政治中的金钱(例如科技界人士的大规模竞选支出)面临着严格的制衡,对结果的影响有限。 ### **具体主题与预测** * **原子与比特(机器人与驾驶):** * 泰勒指出,原子和机器人技术(如Waymo和无人驾驶汽车)进展缓慢。Waymo尚未进入华盛顿特区等城市,它们目前的市场渗透率和对人类福祉的贡献微乎其微。 * **对齐与机制设计:** * 从经济学家的角度使用机制设计来处理对齐问题意味着,智能体不需要共享人类的偏好;可以通过制衡(三权分立)来构建系统,以便自私的智能体能够按照委托人的利益行事。 * 抱脸(Hugging Face)智能体事件是人类错误和激励机制破坏的结果,而不是一场巨大的灾难,这凸显了系统如何通过摩擦进行学习和适应。 * 泰勒强烈不同意“一两年内很可能会发生与人工智能相关的死亡事件”这一说法,称其为歇斯底里。 * **就业与劳动力市场:** * 亚历克斯担心,如果发生协调突破,可能会导致快速转型的潜力,从而引发工资和就业的相变,此时无法进行快速重新匹配,从而在劳动力市场中造成创伤。 * **资本与劳动力(皮凯蒂模型与稀缺性):** * 在讨论多凯什(Dorkesh)和菲尔·特拉梅尔(Phil Trammell)将皮凯蒂模型应用于人工智能世纪的文章时,亚历克斯和泰勒指出,虽然资本将大有作为,但“所有垄断租金都将归资本所有者所有”这一命题尚未得到证实。 * 价值向稀缺性转移。如果能源或特定地点(如太浩湖)依然稀缺,租金就会聚集在那里,尽管人工智能本身可能会使能源或土地改造变得更便宜。 * **未来技术与高等教育:** * 泰勒相信埃里克(Eric)能活到120岁,而他自己则活不到。 * 亚历克斯强调,数字孪生以及人工智能创建微观和宏观现象结构模型(例如细胞模型或宏观经济学模型)的潜力,是一项被严重低估的巨大突破。 * 高等教育预计将在其传统形式上缩减,但人们预计会出现新的实验、衡量标准和市场驱动的创新。芝加哥经济学派关于聪明人管理人工智能智能体和社会科学工具包的愿景被视为未来。

Here is a summary of the podcast, including all news, facts, and opinions presented in the transcript: ### **General Sentiment and Economic Outlook on AI** * **The AI Landscape & Progress:** The conversation took place on November 11th, 2026. Both Alex and Tyler agree that AI progress is incredible and accelerating, marked by frequent and more powerful model releases. * **Chatbot Era vs. Agentic Era:** The shift from the chatbot era (used like a fancy Wikipedia) to the agentic era (where systems perform actual work) has caused a massive acceleration in capabilities and token usage. However, agentic systems use significantly more tokens, and price competition among frontier labs (OpenAI, Anthropic, Muse, DeepMind) has driven token prices down, resulting in flat revenue numbers despite exponential token usage. * **Economic Impact and Growth Rates:** * Tyler and Alex push back against the "San Francisco consensus" of runaway, double-digit GDP growth (singularity-type growth). Based on data and the Ben Jones/Chad Jones/Aghion paper on idea production functions, the parameter for increasing returns is decidedly less than one, pointing toward diminishing returns and a more normal single-digit growth path. * Market data (real interest rates, stock prices), prediction markets for the Anthropic IPO (not flying above $2 trillion), and normal institutional operations indicate a more normal path for the economy. * Real economic transformation requires not just competence, but the ability of agents to coordinate with one another. Currently, agentic swarms are bad at coordinating and mostly do parallel processing. * The U.S. economy contains many massive bottlenecks, inertia, and non-startup sectors (healthcare, higher education, government, nonprofits) that slow down technological diffusion. * Tyler estimates that agents reaching 10% of GDP will not happen soon (though reaching 1.5% within five years is plausible). * **Macro & Political Events:** * The midterms caused Tyler to update on Texas, which surprised him by flipping. * Recent elections and political friction regarding data centers may raise costs and slow things down slightly due to regulation, but will not permanently stall the sector. Money in politics (such as massive campaign spending by tech figures) faces heavy checks and balances and has limited impact on outcomes. ### **Specific Topics and Predictions** * **Atoms vs. Bits (Robotics and Driving):** * Tyler notes that atoms and robotics (like Waymo and driverless cars) are progressing slowly. Waymo has not reached cities like Washington, D.C., yet, and their market penetration and contribution to human welfare are currently tiny. * **Alignment and Mechanism Design:** * Approaching alignment from an economist's perspective using mechanism design implies that agents do not need to share human preferences; systems can be built with checks and balances (separation of powers) so that self-interested agents act in the principal's interest. * The Hugging Face agent incident was the result of human error and broken incentives rather than a massive disaster, highlighting how systems learn and adapt through friction. * Tyler strongly disagrees with the claim that an AI-related death is likely a year or two from now, calling it hysterical. * **Employment and Labor Markets:** * Alex is concerned about the potential for a fast transition if a coordination breakthrough occurs, leading to a phase transition in wages and employment where rapid rematching cannot happen, causing scarring in the labor market. * Tyler notes that employment will face transition problems, search frictions, and a natural rate of unemployment because everyone is currently a novice at using AI, but new projects and jobs will emerge, keeping labor markets fundamentally resilient. * **Capital vs. Labor (Piketty Model & Scarcity):** * Discussing the essay by Dorkesh and Phil Trammell applying Piketty's model to the AI century, Alex and Tyler note that while capital will do a lot, the proposition that all monopoly rents will accrue to capital owners is unproven. * Value moves toward scarcity. If energy or specific locations (like Tahoe) remain scarce, rents will accrue there, though AI itself might make energy or land terraforming cheaper. * **Future Technologies and Higher Education:** * Tyler believes Eric will live to 120, while he will not. * Alex highlights digital twins and AI's potential to create structural models of micro and macro phenomena (such as models of the cell or macroeconomics) as a massive underrated breakthrough. * Higher education is expected to shrink in its traditional form, but new experimentation, measurement, and market-driven innovations are anticipated. The Chicago economic vision of smart people managing AI agents and social science toolkits is seen as the future.

摘要

AI capabilities are improving fast. But will the economy move just as quickly? a16z General Partner Erik Torenberg sits down with Google DeepMind economist Alex Imas and economist Tyler Cowen to examine how advances in AI translate into growth, jobs, wages, and productivity. They debate what it would take to reach double-digit economic growth, why bottlenecks and falling AI costs could slow the impact, and whether frontier lab valuations require much larger productivity gains. They also explore the “economics of intelligence,” including why the bigger shift may come not from smarter individual agents, but from AI agents learning to coordinate at scale. Timestamps: 00:00 Intro 00:58 Is AI progress speeding up? 04:44 Why Tyler is a little more cautious 10:49 Could AI really double economic growth? 19:02 Is GDP the right way to measure AI’s impact? 23:51 What the Hugging Face hack says about AI safety 29:53 Why AI agents still can’t work as a team 33:45 What AI means for jobs and wages 39:29 Elections, data centers and AI politics 43:23 How Alex and Tyler got here 50:00 Rapid fire: AI predictions and higher ed Resources: Follow Alex Imas on X: https://x.com/alexolegimas Follow Tyler Cowen on X: https://x.com/tylercowen Follow Erik Torenberg on X: https://x.com/eriktorenberg Stay Updated: If you enjoyed this episode, be sure to like, subscribe, and share with your friends! Find a16z on X: https://twitter.com/a16z Find a16z on LinkedIn: https://www.linkedin.com/company/a16z Listen to the a16z Show on Spotify: https://open.spotify.com/show/5bC65RDvs3oxnLyqqvkUYX Listen to the a16z Show on Apple Podcasts: https://podcasts.apple.com/us/podcast/a16z-podcast/id842818711 Follow our host: https://x.com/eriktorenberg Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see http://a16z.com/disclosures.

GPT-4正在为你翻译摘要中......

中英文字稿